A Passive Rotational Speed Calculation Method and System for Special Equipment

Abstract

The present invention provides a method and system for calculating the passive rotational speed of special equipment. Among them, the method takes into account the differences in the rotational frequencies of rotating equipment and interference components at different positions of the mechanical body. The multi-channel vibration data with richer information can be jointly used as input parameters, thereby excluding the incoherent frequency components in a single channel. An improved method of second-order statistical blind source separation combined with Vold-Kalman filtering is adopted, which can automatically determine the number of sources to be separated and optimize the multi-time delay input, and accurately identify the instantaneous rotational speed of special equipment. The solution proposed by the present invention can accurately estimate the instantaneous rotational speed in complex environments, such as adjacent frequency interference and high noise levels.

Description

Technical Field

[0001] The present invention belongs to the field of rotational speed extraction of rotating equipment, and particularly relates to a passive rotational speed calculation method and system for special equipment. Background Technique

[0002] Special equipment belongs to complex special equipment with complex operating conditions. To ensure the reliable operation of the equipment, it is necessary to monitor the vibration state of the equipment. However, since special equipment belongs to heavy-duty equipment, the equipment runs at a slow speed, has a short operating duration, and each operation is accompanied by an acceleration and deceleration process, and the steady-state operation time of the equipment is short. Conventional vibration data processing algorithms are mostly used for the processing of steady-state signals. For the processing of non-steady-state vibration signals, in engineering, order tracking analysis of rotating equipment is mostly achieved by increasing synchronous sampling speed. However, due to the limitations of the structure and use scenario of special equipment, a synchronous sampling speed measurement device cannot be installed. Therefore, it is necessary to calculate the rotational speed of special equipment by other methods.

[0003] Many scholars at home and abroad have conducted extensive research on the estimation of rotational speed. Shi Zhensheng et al. used the short-time Fourier transform to transform the vibration signal into the time-frequency space and used the peak extraction method to estimate the instantaneous rotational speed of special equipment; similarly, according to the idea of peak search, Hu Aijun et al. proposed a method for extracting the rotational speed of special equipment based on improved peak search. This method mainly uses the change difference of the first derivative of the positions of two adjacent points near the peak point to be extracted as the judgment factor for distinguishing the true and false peak positions; J., Urbanek et al. proposed a two-step method for extracting the instantaneous rotational speed based on time-frequency analysis. This method first preliminarily obtains the time-frequency peak position points in the time-frequency analysis as a rough estimate of the instantaneous rotational speed, and then demodulates the phase of the original data based on these time-frequency points, and performs time-frequency analysis on the modulated signal again to identify the instantaneous rotational frequency of the machine.

[0004] Most of the existing methods for extracting the rotational speed of rotating equipment only consider the vibration data of a single channel, and generally correct and estimate the rotational frequency by extracting the position of the time-frequency peak ridge through time-frequency transformation. However, in practical problems, there are often multiple adjacent frequencies and noise components in the rotational frequency range to be investigated, and it is often difficult to distinguish these interference information only through the vibration data of a single channel. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes a technical solution for a passive rotational speed calculation method and system for special equipment to solve the above technical problems.

[0006] The first aspect of the present invention discloses a passive rotational speed calculation method for special equipment, and the method includes:

[0007] Step S1, obtain vibration signals of multiple channels, and determine the rotational frequency range according to the special equipment model;

[0008] Step S2: Perform preprocessing operations of filtering, resampling, and intensity normalization on the vibration signal according to the rotation frequency range to obtain a preprocessed vibration signal;

[0009] Step S3: Calculate the covariance matrix of the preprocessed vibration signal at time delay of 0;

[0010] Step S4: Calculate the eigenvalues and corresponding eigenvectors of the covariance matrix of the preprocessed vibration signal at time delay of 0, arrange the eigenvalues in descending order, retain the first n eigenvalues whose values are greater than a predetermined proportion of the total sum, and use the remaining m - n eigenvalues to calculate the estimated value of the vibration signal noise level, where m is the total number of eigenvalues of the covariance matrix of the preprocessed vibration signal;

[0011] Step S5: Calculate the whitening matrix of the vibration signal by applying the estimated value of the vibration signal noise level, the first n eigenvalues, and the corresponding eigenvectors;

[0012] Step S6: Multiply the whitening matrix with the preprocessed vibration signal to obtain a whitened vibration signal, and calculate the covariance matrix of the whitened vibration signal at K random time delays τ;

[0013] Step S7: Perform joint diagonalization on the covariance matrix of the whitened vibration signal to separate the obtained source signal sequence;

[0014] Step S8: According to the spectral characteristics of the multi-channel vibration data, select the source signals within the rotation frequency range from the source signal sequence, and perform Hilbert transform on the source signals to calculate the instantaneous frequency at any time of the source signals;

[0015] Step S9: Perform Vold-Kalman filtering with a random bandwidth on the source signals according to the instantaneous frequency to obtain a corrected time-domain waveform of the target rotation frequency;

[0016] Step S10: Apply Hilbert transform to extract the instantaneous frequency of the time-domain waveform to obtain the rotation frequency of the target;

[0017] Step S11: Repeat and iterate Step S9 and Step S10, calculate the rotation frequency iteration error by using the instantaneous frequency obtained in each iteration and the rotation frequency of the target. When the rotation frequency iteration error is less than a preset threshold, it is considered that the model converges, stop the iteration, and calculate the estimated value of the instantaneous rotational speed by using the rotation frequency of the target in the last iteration.

[0018] According to the method of the first aspect of the present invention, in the step S4, the method for calculating the estimated value of the vibration signal noise level includes:

[0019] When n < m,

[0020] When n = m,

[0021] where, is the estimated value of the vibration signal noise level, and λ i is the eigenvalue.

[0022] According to the method of the first aspect of the present invention, in the step S5, the method for calculating the whitening matrix of the vibration signal includes:

[0023]

[0024] where the symbol [·] H represents the Hermitian transpose of the matrix, and v1…v n are the eigenvectors corresponding to the eigenvalues λ1…λ n , is the estimated value of the vibration signal noise level, and W is the whitening matrix.

[0025] According to the method of the first aspect of the present invention, in the step S7, the method for jointly diagonalizing the covariance matrix of the whitened vibration signal and separating the obtained source signal sequence includes:

[0026] Calculate the orthogonal matrix U such that {R zz (τ) K = U{D} K U H , where {D} K is the K pair of focusing matrices;

[0027] The separated source signal sequence Z is the whitened vibration signal.

[0028] According to the method of the first aspect of the present invention, in the step S8, the method for calculating the instantaneous frequency of any moment of the source signal by performing the Hilbert transform on the source signal includes:

[0029]

[0030]

[0031] where H(·) is the Hilbert transform, is the source signal selected from the source signal sequence within the rotation frequency range, Θ(t) is the calculated intermediate variable, and F inst (t) is the instantaneous frequency.

[0032] According to the method of the first aspect of the present invention, in the step S11, the method for calculating the rotational frequency iteration error based on the instantaneous frequency and the rotational frequency of the target obtained in each iteration of the application includes:

[0033]

[0034] where error is the rotational frequency iteration error, F inst,j (t) is the instantaneous frequency, F′ inst (t) is the rotational frequency of the target, and L is the length of the iterated data.

[0035] According to the method of the first aspect of the present invention, in the step S11, the method for calculating the estimated value of the instantaneous rotational speed includes:

[0036] N(t) = 60 × F′ inst ()

[0037] where N(t) is the estimated value of the instantaneous rotational speed, and F′ inst (t) is the rotational frequency of the target.

[0038] The second aspect of the present invention discloses a passive rotational speed calculation system for special equipment, the system includes:

[0039] A first processing module, configured to obtain multi-channel vibration signals and determine the rotational frequency range according to the special equipment model;

[0040] A second processing module, configured to perform preprocessing operations of filtering, resampling, and intensity normalization on the vibration signals according to the rotational frequency range to obtain preprocessed vibration signals;

[0041] A third processing module, configured to calculate the covariance matrix of the preprocessed vibration signals at time delay 0;

[0042] A fourth processing module, configured to calculate the eigenvalues and corresponding eigenvectors of the covariance matrix of the preprocessed vibration signals at time delay 0, arrange the eigenvalues in descending order, retain the first n eigenvalues whose eigenvalues are greater than a predetermined proportion of the total sum, and the remaining m - n eigenvalues are used to calculate the estimated value of the vibration signal noise level, where m is the total number of eigenvalues of the covariance matrix of the preprocessed vibration signals;

[0043] A fifth processing module, configured to calculate the whitening matrix of the vibration signals by applying the estimated value of the vibration signal noise level, the first n eigenvalues, and the corresponding eigenvectors;

[0044] The sixth processing module is configured to multiply the whitening matrix with the preprocessed vibration signal to obtain the whitened vibration signal, and calculate the covariance matrix of the whitened vibration signal at K random time delays τ;

[0045] The seventh processing module is configured to perform joint diagonalization on the covariance matrix of the whitened vibration signal to separate the obtained source signal sequence;

[0046] The eighth processing module is configured to select, according to the spectral characteristics of the multi-channel vibration data, the source signals within the rotating frequency range from the source signal sequence, and perform Hilbert transform on the source signals to calculate the instantaneous frequency at any moment of the source signals;

[0047] The ninth processing module is configured to perform Vold-Kalman filtering with a random bandwidth on the source signals according to the instantaneous frequency to obtain the corrected time-domain waveform of the target rotating frequency;

[0048] The tenth processing module is configured to apply Hilbert transform to extract the instantaneous frequency of the time-domain waveform to obtain the target rotating frequency;

[0049] The eleventh processing module is configured to repeat the ninth processing module and the tenth processing module, calculate the rotating frequency iteration error by using the instantaneous frequency obtained each time and the target rotating frequency. When the rotating frequency iteration error is less than a preset threshold, it is considered that the model converges, the iteration is stopped, and the instantaneous rotational speed estimation value is calculated by using the target rotating frequency of the last iteration.

[0050] For the system according to the second aspect of the present invention, the fourth processing module is configured to, the calculation of the vibration signal noise level estimation value includes:

[0051] When n < m,

[0052] When n = m,

[0053] Wherein, is the vibration signal noise level estimation value, and λ i is the eigenvalue.

[0054] For the system according to the second aspect of the present invention, the fifth processing module is configured to, the method for calculating the whitening matrix of the vibration signal includes:

[0055]

[0056] Wherein, the symbol [·] H represents the Hermitian transpose of the matrix, and v1... v n are the eigenvectors corresponding to the eigenvalues λ1... λ n ; is the estimated value of the vibration signal noise level, and W is the whitening matrix.

[0057] According to the system of the second aspect of the present invention, a seventh processing module is configured to jointly diagonalize the covariance matrix of the whitened vibration signal, and the separated source signal sequence includes:

[0058] Calculate the orthogonal matrix U such that {R zz (τ)} K = U{D} K U H , where {D} K is the K pair of focusing matrices;

[0059] The separated source signal sequence Z is the whitened vibration signal.

[0060] According to the system of the second aspect of the present invention, an eighth processing module is configured to calculate the instantaneous frequency at any time of the source signal by performing a Hilbert transform on the source signal, including:

[0061]

[0062]

[0063] where H(·) is the Hilbert transform, is the source signal selected from the source signal sequence within the rotation frequency range, Θ(t) is the calculated intermediate variable, and F inst (t) is the instantaneous frequency.

[0064] According to the system of the second aspect of the present invention, an eleventh processing module is configured to calculate the rotation frequency iteration error by applying the instantaneous frequency obtained in each iteration and the target rotation frequency, including:

[0065]

[0066] where error is the rotation frequency iteration error, F inst,j (t) is the instantaneous frequency, F′ inst (t) is the target rotation frequency, and L is the data length of the iteration.

[0067] According to the system of the second aspect of the present invention, an eleventh processing module is configured to calculate the instantaneous rotational speed estimate, including:

[0068] N(t) = 60 × F′ inst (t)

[0069] where N(t) is the instantaneous rotational speed estimate, and F′ inst (t) is the target rotation frequency.

[0070] A third aspect of the present invention discloses an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the steps in a passive rotational speed calculation method for a special equipment according to any one of the first aspects of the present disclosure are implemented.

[0071] A fourth aspect of the present invention discloses a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the steps in a passive rotational speed calculation method for a special equipment according to any one of the first aspects of the present disclosure are implemented.

[0072] The solution proposed by the present invention can accurately estimate the instantaneous rotational speed in complex environments, such as adjacent frequency interference and high noise levels. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0074] Figure 1 FIG. is a flowchart of a passive rotational speed calculation method for a special equipment according to an embodiment of the present invention;

[0075] Figure 2 FIG. is the acceleration data of a certain motor at 4 measurement points according to an embodiment of the present invention;

[0076] Figure 3 FIG. is the frequency spectrum of the acceleration data according to an embodiment of the present invention;

[0077] Figure 4 FIG. is the whitened vibration data according to an embodiment of the present invention;

[0078] Figure 5 FIG. is the separated source signal and its spectrogram according to an embodiment of the present invention;

[0079] Figure 6 FIG. is the time-domain waveform of the source signal 1 after being corrected by Vold-Kalman filtering according to an embodiment of the present invention;

[0080] Figure 7 FIG. is the convergence of the error during the iteration process according to an embodiment of the present invention;

[0081] Figure 8 FIG. is the estimated instantaneous rotational speed according to an embodiment of the present invention;

[0082] Figure 9 It is a structural diagram of a passive rotational speed calculation system for special equipment according to an embodiment of the present invention;

[0083] Figure 10 It is a structural diagram of an electronic device according to an embodiment of the present invention. Specific embodiments

[0084] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0085] The present invention aims to disclose a new method for jointly extracting the rotational speed of special equipment through multi-channel vibration signals, which can accurately estimate the instantaneous rotational speed in complex environments (such as adjacent frequency interference and high noise levels). This method uses a method of blind source separation combined with Vold-Kalman filtering. Different from the common rotational speed estimation methods based on single-channel data, this method takes into account the consistency of the rotational characteristic frequencies of rotating equipment at different positions of the mechanical body. The multi-channel vibration data can be jointly used as input parameters. Therefore, this method is not easily affected by local frequency components and has high recognition accuracy and robustness. The blind source separation (BSS) technology is a class of unsupervised learning methods in machine learning, and its goal is to reconstruct the statistically uncorrelated source components and their mixing coefficient matrix in the signal only based on the observed signals. The rationality of applying the BSS algorithm to the rotational speed extraction of special equipment is that the rotational frequency exists in different sensors and is independent of other frequency components. Since the reciprocity of the rotational frequency time-domain signal to be separated is good and the local correlation is obvious, the separation operation can be carried out starting from the high-order statistics of the vibration signal. However, in some complex situations, the separated rotational frequency time-domain source signal may still be affected by noise and weak adjacent frequency components. At the same time, the BSS algorithm itself still has some limitations, such as the uncertainty of the number of signal sources and the serious influence of the time delay on the robustness of the separation result, which can be further corrected. Therefore, in this solution, an improved second-order statistics blind source separation combined with Vold-Kalman filtering method is adopted, which can automatically determine the number of sources to be separated and optimize the multi-time delay input. Vold-Kalman filtering is an order tracking analysis algorithm commonly used for special equipment, but its bandwidth selection often affects the quality of the final filtering. Therefore, in the present invention, through an iterative idea, the separated rotational frequency time-domain signal is subjected to Vold-Kalman filtering correction multiple times, and the non-target frequency components and test noise hidden in the vibration signal are effectively eliminated, and then the instantaneous rotational speed of special equipment can be accurately identified.

[0086] Example 1:

[0087] The first aspect of the present invention discloses a method for calculating the passive rotational speed of special equipment. Figure 1 As shown in the flowchart of a method for calculating the passive rotational speed of special equipment according to an embodiment of the present invention, Figure 1 the method includes:

[0088] Step S1, obtain multi-channel vibration signals, and estimate the rotational frequency range according to the special equipment model;

[0089] Step S2, perform preprocessing operations of filtering, resampling, and intensity normalization on the vibration signals according to the rotational frequency range to obtain preprocessed vibration signals;

[0090] Step S3, calculate the covariance matrix at time delay 0 of the preprocessed vibration signals;

[0091] Step S4, calculate the eigenvalues and corresponding eigenvectors of the covariance matrix of the preprocessed vibration signals, arrange the eigenvalues in descending order, retain the first n eigenvalues whose eigenvalues are greater than a predetermined proportion of the total sum, and the remaining m - n eigenvalues are used to calculate the estimated value of the vibration signal noise level, where m is the total number of eigenvalues of the covariance matrix of the preprocessed vibration signals;

[0092] Step S5, calculate the whitening matrix of the vibration signals by applying the estimated value of the vibration signal noise level, the first n eigenvalues, and the corresponding eigenvectors;

[0093] Step S6, multiply the whitening matrix with the preprocessed vibration signals to obtain whitened vibration signals, and calculate the covariance matrix of the whitened vibration signals at K random time delays τ;

[0094] Step S7, perform joint diagonalization on the covariance matrix of the whitened vibration signals to separate the obtained source signal sequence;

[0095] Step S8, according to the spectral characteristics of the multi-channel vibration data, select the source signals within the rotational frequency range from the source signal sequence, and perform Hilbert transform on the source signals to calculate the instantaneous frequency at any time of the source signals;

[0096] Step S9, perform Vold-Kalman filtering with a random bandwidth on the source signals according to the instantaneous frequency to obtain the time-domain waveform of the corrected target rotational frequency;

[0097] Step S10, apply Hilbert transform to extract the instantaneous frequency of the time-domain waveform to obtain the rotational frequency of the target;

[0098] Step S11: Repeatedly iterate steps S9 and S10, calculate the rotational frequency iteration error using the instantaneous frequency obtained in each iteration and the target rotational frequency. When the rotational frequency iteration error is less than the preset threshold, it is considered that the model converges, stop the iteration, and use the target rotational frequency of the last iteration to calculate the estimated value of the instantaneous rotational speed.

[0099] In step S1, obtain multi-channel vibration signals and estimate the rotational frequency range for the special equipment model.

[0100] In step S2, perform preprocessing operations of filtering, resampling, and intensity normalization on the vibration signals according to the rotational frequency range to obtain the preprocessed vibration signal X = [x1,..x m T , where the symbol (·) T represents the transpose operation of the matrix, and m is the number of channels of the vibration signal.

[0101] In step S3, calculate the covariance matrix R xx (0) at the time delay τ = 0 of the preprocessed vibration signal.

[0102] In step S4, calculate the eigenvalues λ and the corresponding eigenvectors υ of the covariance matrix R xx (0) of the preprocessed vibration signal, and arrange the eigenvalues in descending order. Retain the first n (n ≤ m) eigenvalues λ1,…λ n whose eigenvalues are greater than 5% of the total sum, and the remaining m - n eigenvalues λ n+1 …λ m are used to calculate the estimated value of the vibration signal noise level The m is the total number of eigenvalues of the covariance matrix of the preprocessed vibration signal.

[0103] In some embodiments, in step S4, the method for calculating the estimated value of the vibration signal noise level includes:

[0104] When n < m,

[0105] When n = m, it belongs to positive definite blind source separation. At this time, it is considered that the input information is not sufficient to support the estimation of the noise intensity.

[0106] Among them, is the estimated value of the vibration signal noise level, and λ i is the eigenvalue.

[0107] In step S5, calculate the whitening matrix W of the vibration signal using the estimated value of the vibration signal noise level, the first n eigenvalues, and the corresponding eigenvectors.

[0108] ​In some embodiments, in the step S5, the method for calculating the whitening matrix of the vibration signal includes:

[0109]

[0110] where the symbol (·) H represents the Hermitian transpose of the matrix, and v1…v n are the eigenvectors corresponding to the eigenvalues λ1…λ n , is the estimated value of the vibration signal noise level, and W is the whitening matrix.

[0111] In step S6, multiply the whitening matrix W by the preprocessed vibration signal X to obtain the whitened vibration signal Z = WX, and calculate the covariance matrix R zz (τ) of the whitened vibration signal at K random time delays τ.

[0112] In step S7, perform joint diagonalization on the covariance matrix R zz (τ) of the whitened vibration signal to separate the obtained source signal sequence

[0113] In some embodiments, in the step S7, the method for performing joint diagonalization on the covariance matrix of the whitened vibration signal to separate the obtained source signal sequence includes:

[0114] Calculate the orthogonal matrix U such that {R zz (τ)} K = U{D} K U H , where {D} K is the K pair of focusing matrices, and (·) H represents the Hermitian transpose of the matrix;

[0115] Separate the obtained source signal sequence Z is the whitened vibration signal, and the mixing matrix where the symbol (·) # is the operation of finding the generalized inverse matrix.

[0116] In step S8, according to the spectral characteristics of the multi-channel vibration data, select the source signals within the running frequency range from the source signal sequence, and perform Hilbert transform on the source signals to calculate the instantaneous frequency F inst (t) at any time of the source signals.

[0117] In some embodiments, in the step S8, the method for performing Hilbert transform on the source signals to calculate the instantaneous frequency at any time of the source signals includes:

[0118]

[0119]

[0120] Among them, H(·) is the Hilbert transform, is the source signal within the rotating frequency range selected from the source signal sequence, Θ(t) is the calculated intermediate variable, and F inst (t) is the instantaneous frequency.

[0121] In step S9, according to the instantaneous frequency F inst,j (t), the source signal is subjected to Vold-Kalman filtering with a random bandwidth to obtain the corrected time-domain waveform of the target rotating frequency

[0122] In step S10, the Hilbert transform is applied to extract the instantaneous frequency of the time-domain waveform to obtain the target rotating frequency F′ inst (t).

[0123] In step S11, steps S9 and S10 are repeatedly iterated. The rotating frequency iteration error is calculated using the instantaneous frequency and the target rotating frequency obtained in each iteration. When the rotating frequency iteration error is less than 5%, the model is considered to have converged, the iteration is stopped, and the target rotating frequency of the last iteration is applied to calculate the instantaneous rotational speed estimation value.

[0124] In some embodiments, in step S11, the method of calculating the rotating frequency iteration error using the instantaneous frequency and the target rotating frequency obtained in each iteration includes:

[0125]

[0126] where error is the rotating frequency iteration error, F inst,j (t) is the instantaneous frequency, F′ inst (t) is the target rotating frequency, and L is the data length of the iteration.

[0127] The method of calculating the instantaneous rotational speed estimation value includes:

[0128] N(t) = 60 × F′ inst (t)

[0129] where N(t) is the instantaneous rotational speed estimation value and F′ inst (t) is the target rotating frequency.

[0130] Embodiment 2:

[0131] As Figure 2As shown, the vibration data of four different measuring points collected from a certain motor during operation. The installed sensing system continuously collects the acceleration data of the equipment at a fixed sampling frequency (2560 Hz). Now, data with a length L = 8192 is intercepted for rotational speed identification. According to the model of this motor and its historical operating conditions, the rotational frequency range is known to be 0 Hz to 60 Hz.

[0132] Figure 3 It is the spectrogram of the vibration data of this motor in the low-frequency section. It can be found that the rotational frequency of this motor is severely interfered by other adjacent vibration frequency components, and the interference frequencies of different channels are different, showing complex characteristics. For the data of a single channel, it is not only difficult to distinguish the rotational frequency, but also very difficult to eliminate the interference of adjacent frequencies on the rotational frequency identification.

[0133] Figure 4 It is Figure 2 the result after the preprocessing in step 2 and whitening in steps 3 to 5 of the vibration data. It should be noted that in the calculation of step 4, it is found that the eigenvalues of the covariance matrix of the vibration data at zero delay are λ = [0.0341, 0.0234, 0.0064, 0.0021]. At this time, the eigenvalues to be retained should be greater than 0.0034. Therefore, the number of source signals is estimated to be 3, and the estimated value of the noise intensity is 0.0021.

[0134] Figure 5 It is Figure 4 the source signals and the corresponding spectrogram obtained after the joint diagonalization operation on the whitened vibration data of Figure 5 . It can be seen that through the present invention, different frequency components in multiple channels have been clearly distinguished. Through the spectral analysis of the source signals, it can be known that the first source signal exists in all four sensors participating in the calculation. Based on this, it can be known that source signal 1 is the time-domain signal of the rotational frequency. Through

[0135] Figure 6 It is Figure 5 the result of performing Vold-Kalman filtering on the time-domain waveform of source signal 1 in From Figure 6 it can be clearly found that through the present invention, the multi-channel vibration signals have been extracted into a time-domain waveform with a single frequency component of the rotational frequency.

[0136] Figure 7 It is the change of the rotational frequency error value with the number of iterations during the iterative process. As can be seen from the figure, in this example, the iterative error decreases with the increase of the number of iterations and finally converges to about 3%. The preset error threshold of 5% can meet the requirements of the measured data.

[0137] Figure 8 Comparison between the instantaneous rotational speed data identified by this method and the actual data.

[0138] In summary, the solution proposed by the present invention can accurately estimate the instantaneous rotational speed in complex environments, such as adjacent frequency interference and high noise levels.

[0139] Embodiment 3:

[0140] Embodiment 3 discloses a passive rotational speed calculation system for special equipment. Figure 9 FIG. is a structural diagram of a passive rotational speed calculation system for special equipment according to an embodiment of the present invention; as Figure 9 shown, the system 100 includes:

[0141] A first processing module 101, configured to obtain multi-channel vibration signals and estimate the rotational frequency range for the special equipment model;

[0142] A second processing module 102, configured to perform preprocessing operations of filtering, resampling, and intensity normalization on the vibration signals according to the rotational frequency range to obtain preprocessed vibration signals;

[0143] A third processing module 103, configured to calculate the covariance matrix of the preprocessed vibration signals at time delay 0;

[0144] A fourth processing module 104, configured to calculate the eigenvalues and corresponding eigenvectors of the covariance matrix of the preprocessed vibration signals, arrange the eigenvalues in descending order, retain the first n eigenvalues whose eigenvalues are greater than a predetermined proportion of the total sum, and use the remaining m - n eigenvalues to calculate the vibration signal noise level estimate value, where m is the total number of eigenvalues of the covariance matrix of the preprocessed vibration signals;

[0145] A fifth processing module 105, configured to calculate the whitening matrix of the vibration signals by applying the vibration signal noise level estimate value, the first n eigenvalues, and the corresponding eigenvectors;

[0146] A sixth processing module 106, configured to multiply the whitening matrix by the preprocessed vibration signals to obtain whitened vibration signals, and calculate the covariance matrix of the whitened vibration signals at K random time delays τ;

[0147] A seventh processing module 107, configured to perform joint diagonalization on the covariance matrix of the whitened vibration signals to separate the obtained source signal sequence;

[0148] The eighth processing module 108 is configured to select source signals within the rotational frequency range from the source signal sequence according to the spectral characteristics of the multi-channel vibration data, and perform Hilbert transform on the source signals to calculate the instantaneous frequency of the source signals at any moment;

[0149] The ninth processing module 109 is configured to perform Vold-Kalman filtering with a random bandwidth on the source signals according to the instantaneous frequency to obtain the time-domain waveform of the corrected target rotational frequency;

[0150] The tenth processing module 1010 is configured to apply Hilbert transform to extract the instantaneous frequency of the time-domain waveform to obtain the target rotational frequency;

[0151] The eleventh processing module 1011 is configured to repeat the ninth processing module and the tenth processing module, calculate the rotational frequency iteration error by using the instantaneous frequency and the target rotational frequency obtained from each repetition. When the rotational frequency iteration error is less than a preset threshold, it is considered that the model converges, the iteration is stopped, and the instantaneous rotational speed estimation value is calculated by using the target rotational frequency of the last iteration.

[0152] Furthermore, the fourth processing module 104 is configured to: the calculation of the vibration signal noise level estimation value includes:

[0153] When n < m,

[0154] When n = m,

[0155] Wherein, is the vibration signal noise level estimation value, and λ i is the eigenvalue.

[0156] Specifically, the fifth processing module 105 is configured to: the method for calculating the whitening matrix of the vibration signal includes:

[0157]

[0158] Wherein, the symbol (·) H represents the Hermitian transpose of the matrix, and v1...v n are the eigenvectors corresponding to the eigenvalues λ1...λ n respectively, is the vibration signal noise level estimation value, and W is the whitening matrix.

[0159] Wherein, the seventh processing module 107 is configured to: the calculation of the whitening matrix of the vibration signal includes:

[0160]

[0161] Wherein, the symbol (·)H represents the Hermitian transpose of the matrix, v1…v n are the eigenvalues λ1…λ n corresponding eigenvectors, is the estimated value of the vibration signal noise level, and W is the whitening matrix.

[0162] Furthermore, the seventh processing module 107 is configured to jointly diagonalize the covariance matrix of the whitened vibration signal, and the separated source signal sequence includes:

[0163] Calculate the orthogonal matrix U such that {R zz (τ)} K = U{D} K U H , where {D} K is the K pair of focusing matrices, (·) H represents the Hermitian transpose of the matrix;

[0164] The separated source signal sequence Z is the whitened vibration signal.

[0165] Preferably, the eighth processing module is configured to calculate the instantaneous frequency of any moment of the source signal by performing a Hilbert transform on the source signal, including:

[0166]

[0167]

[0168] where H(·) is the Hilbert transform, is the source signal selected from the source signal sequence within the rotating frequency range, Θ(t) is the calculated intermediate variable, and F inst (t) is the instantaneous frequency.

[0169] The eleventh processing module 1011 is configured to calculate the rotating frequency iteration error by applying the instantaneous frequency obtained in each iteration and the target rotating frequency, including:

[0170]

[0171] where error is the rotating frequency iteration error, F inst,j (t) is the instantaneous frequency, F′ inst (t) is the target rotating frequency, and L is the length of the iterative data.

[0172] The eleventh processing module 1011 is configured to calculate the estimated value of the instantaneous rotational speed, including:

[0173] N(t) = 60 × F′ inst (t)

[0174] Among them, N(t) is the estimated value of the instantaneous rotational speed, and F' inst (t) is the rotational frequency of the target.

[0175] Embodiment 4:

[0176] Embodiment 4 of the present disclosure discloses an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the steps in a passive rotational speed calculation method for special equipment according to any one of the first aspects disclosed in the present invention are implemented.

[0177] Figure 10 FIG. is a structural diagram of an electronic device according to an embodiment of the present invention. As Figure 10 shown, the electronic device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the electronic device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be implemented through WIFI, a carrier network, near field communication (NFC), or other technologies. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen. The input device of the electronic device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the electronic device, or an external keyboard, a touchpad, or a mouse, etc.

[0178] Those skilled in the art can understand that Figure 10 the structure shown in is only a structural diagram of a part related to the technical solution of the present disclosure, and does not constitute a limitation on the electronic device to which the solution of the present application is applied. A specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0179] The fourth aspect of the present invention discloses a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the steps in a passive rotational speed calculation method for special equipment according to any one of the first aspects disclosed in the present invention are implemented.

[0180] Please note that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity in description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as falling within the scope described in this specification. The above embodiments only represent several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A method for calculating the passive rotational speed of special equipment, characterized in that, The method includes: Step S1: Obtain vibration signals of multiple channels, and determine the rotation frequency range for the special equipment model; Step S2: Perform preprocessing operations of filtering, resampling, and intensity normalization on the vibration signals according to the rotation frequency range to obtain preprocessed vibration signals; Step S3: Calculate the covariance matrix of the preprocessed vibration signals at the time delay of 0; Step S4: Calculate the eigenvalues and corresponding eigenvectors of the covariance matrix of the preprocessed vibration signals at the time delay of 0, arrange the eigenvalues in descending order, retain the first n eigenvalues whose eigenvalues are greater than a predetermined proportion of the total sum, and the remaining m - n eigenvalues are used to calculate the estimated value of the vibration signal noise level, where m is the total number of eigenvalues of the covariance matrix of the preprocessed vibration signals; Among them, the method for calculating the estimated value of the vibration signal noise level specifically includes: When n < m, When n = m, Among them, is the estimated value of the vibration signal noise level, and λ i is the i-th eigenvalue; Step S5: Calculate the whitening matrix of the vibration signal based on the estimated value of the vibration signal noise level, the first n eigenvalues, and the corresponding eigenvectors; Step S6: Multiply the whitening matrix by the preprocessed vibration signals to obtain whitened vibration signals, and calculate the covariance matrix of the whitened vibration signals at K random time delays τ; Step S7: Jointly diagonalize the covariance matrix of the whitened vibration signals to separate the source signal sequence; Step S8: According to the spectral characteristics of the multi-channel vibration data, select the source signals within the rotation frequency range from the source signal sequence, and perform Hilbert transform on the source signals to calculate the instantaneous frequency at any time of the source signals; Step S9: Perform Vold-Kalman filtering with a random bandwidth on the source signals according to the instantaneous frequency to obtain the corrected time-domain waveform of the target rotation frequency; Step S10: Use Hilbert transform to extract the instantaneous frequency of the time-domain waveform to obtain the target rotation frequency; Step S11: Repeat and iterate Step S9 and Step S10, calculate the rotation frequency iteration error using the instantaneous frequency and the target rotation frequency obtained in each iteration. When the rotation frequency iteration error is less than the preset threshold, it is considered that the model converges, stop the iteration, and use the target rotation frequency of the last iteration to calculate the estimated value of the instantaneous rotational speed.

2. The method for calculating the passive rotational speed of special equipment according to claim 1, characterized in that, In Step S5, the method for calculating the whitening matrix of the vibration signal includes: where the symbol [·] H represents the Hermitian transpose of the matrix, and v1…v n are the eigenvectors corresponding to the eigenvalues λ1…λ n , is the estimated value of the vibration signal noise level, and W is the whitening matrix.

3. The method for calculating the passive rotational speed of special equipment according to claim 1, characterized in that, In Step S7, the method for jointly diagonalizing the covariance matrix of the whitened vibration signals to separate the source signal sequence includes: Calculate the orthogonal matrix U such that {R zz (τ)} K = U{D} K U H , where {D} K is the K pairs of focus matrices; The separated source signal sequence Z is the vibration signal after whitening.

4. The method for calculating the passive rotational speed of special equipment according to claim 1, characterized in that, In Step S8, the method for performing Hilbert transform on the source signals to calculate the instantaneous frequency at any time of the source signals includes: where \(H(\cdot)\) is the Hilbert transform, is the source signal selected from the source signal sequence within the rotating frequency range, \(\Theta(t)\) is the calculated intermediate variable, \(F\) inst (t) is the instantaneous frequency.

5. The method for calculating the passive rotational speed of special equipment according to claim 1, characterized in that, In Step S11, the method for calculating the rotation frequency iteration error using the instantaneous frequency and the target rotation frequency obtained in each iteration includes: where error is the rotational frequency iteration error, F inst,j (t) is the instantaneous frequency, F' inst (t) is the rotational frequency of the target, and L is the length of the iterative data.

6. The method for calculating the passive rotational speed of special equipment according to claim 5, characterized in that, In Step S11, the method for calculating the estimated value of the instantaneous rotational speed includes: N(t) = 60 × F′ inst (t) where N(t) is the estimated value of the instantaneous rotational speed, and F' inst (t) is the rotational frequency of the target.

7. A passive rotational speed calculation system for special equipment, characterized in that, The system includes: A first processing module, configured to obtain vibration signals of multiple channels, and determine the rotation frequency range for the special equipment model; A second processing module, configured to perform preprocessing operations of filtering resampling and intensity normalization on the vibration signal according to the conversion frequency range to obtain a preprocessed vibration signal; A third processing module, configured to calculate a covariance matrix of the preprocessed vibration signal at the time delay of 0; A fourth processing module, configured to calculate eigenvalues and corresponding eigenvectors of the covariance matrix of the preprocessed vibration signal, arrange the eigenvalues in descending order, retain the first n eigenvalues whose eigenvalues are greater than a predetermined proportion of the total sum, and the remaining m - n eigenvalues are used to calculate an estimated value of the vibration signal noise level, where m is the total number of eigenvalues of the covariance matrix of the preprocessed vibration signal; Wherein, the method for calculating the estimated value of the vibration signal noise level specifically includes: When n < m, When n = m, Among them, is the estimated value of the vibration signal noise level, and λ i is the i-th eigenvalue; A fifth processing module, configured to calculate a whitening matrix of the vibration signal by applying the estimated value of the vibration signal noise level, the first n eigenvalues and the corresponding eigenvectors; A sixth processing module, configured to multiply the whitening matrix by the preprocessed vibration signal to obtain a whitened vibration signal, and calculate a covariance matrix of the whitened vibration signal at K random time delays τ; A seventh processing module, configured to perform joint diagonalization on the covariance matrix of the whitened vibration signal to separate the obtained source signal sequence; An eighth processing module, configured to select a source signal within the conversion frequency range from the source signal sequence according to the spectral characteristics of the multi-channel vibration data, and perform Hilbert transform on the source signal to calculate the instantaneous frequency of the source signal at any time; A ninth processing module, configured to perform Vold-Kalman filtering with a random bandwidth on the source signal according to the instantaneous frequency to obtain a corrected time domain waveform of the target conversion frequency; A tenth processing module, configured to extract the instantaneous frequency of the time domain waveform by applying Hilbert transform to obtain the target conversion frequency; An eleventh processing module, configured to repeat the ninth processing module and the tenth processing module, calculate a conversion frequency iteration error by applying the instantaneous frequency and the target conversion frequency obtained from each repetition, and when the conversion frequency iteration error is less than a preset threshold, it is considered that the model converges, stop the iteration, and calculate an instantaneous rotational speed estimated value by applying the target conversion frequency of the last iteration.

8. An electronic device, characterized in that, The electronic device includes a memory and a processor. When the processor executes a computer program stored in the memory, the steps in any one of claims 1 to 6 of a method for calculating the passive rotational speed of special equipment are implemented.

9. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the steps in any one of claims 1 to 6 of a method for calculating the passive rotational speed of special equipment are implemented.

Images